基于改进GM(1,1)与WOA-LSSVM组合预测模型的轨道不平顺预测
|
摘要
在保障列车行车安全的前提下对轨道不平顺的发展趋势进行预测,可以提高线路维护效率。根据轨检车的历史轨检TQI数值进行分析,提出一种基于非等时距近似非齐次的GM(1,1)模型与鲸鱼算法优化的最小二乘支持向量机的组合预测模型。对非等时距GM(1,1)模型的灰作用量进行优化,并设置加权矩阵,对不同检测时间的数据赋予不同权值,建立非等时距近似非齐次的GM(1,1)模型,得到初步预测值。在此基础上,利用鲸鱼算法优化的最小二乘支持向量机(WOA-LSSVM)对残差进行修正,得到最终预测值。分别对某线上行两段线路的轨道不平顺TQI值进行预测,结果表明:该预测方法相对误差平均值分别为2. 316%和1. 67%,后验差分别为0. 093和0. 068,精度等级达到1级,实现了轨道不平顺较高精度的预测。
Under the premise of protecting the safety of train traffic,effective prediction of irregularity development trend can improve the efficiency of line maintenance. According to the analysis of historical track quality data from track inspection car,this paper proposes a combined forecasting model based on the non-equal interval and non-homogeneous GM( 1,1) model of the least squares support vector machine optimized by whale optimization algorithm. The grey effect amount of Non-equal interval GM( 1,1) model is optimized and a weighting matrix is set up to assign different weights to data obtained at different detection times. Non-equal interval and non-homogeneous GM( 1,1) model is established to get the preliminary prediction value. Based on this,the residual error is corrected by using the least squares support vector machine( WOA-LSSVM) optimized by whale optimization algorithm, and the final prediction value is obtained. The track irregularity TQI values of two up-direction sections on a certain line are predicted,and the results show that the averages of the relative errors of this prediction method are 2. 316% and 1. 67% respectively,the posterior error ratios are 0. 093 3 and 0. 068 1 respectively,and the accuracy grade reaches level 1. Thus,accurate prediction of track irregularity is fulfilled
Under the premise of protecting the safety of train traffic,effective prediction of irregularity development trend can improve the efficiency of line maintenance. According to the analysis of historical track quality data from track inspection car,this paper proposes a combined forecasting model based on the non-equal interval and non-homogeneous GM( 1,1) model of the least squares support vector machine optimized by whale optimization algorithm. The grey effect amount of Non-equal interval GM( 1,1) model is optimized and a weighting matrix is set up to assign different weights to data obtained at different detection times. Non-equal interval and non-homogeneous GM( 1,1) model is established to get the preliminary prediction value. Based on this,the residual error is corrected by using the least squares support vector machine( WOA-LSSVM) optimized by whale optimization algorithm, and the final prediction value is obtained. The track irregularity TQI values of two up-direction sections on a certain line are predicted,and the results show that the averages of the relative errors of this prediction method are 2. 316% and 1. 67% respectively,the posterior error ratios are 0. 093 3 and 0. 068 1 respectively,and the accuracy grade reaches level 1. Thus,accurate prediction of track irregularity is fulfilled
引文
[1]张念,朱焱,王小敏,等.铁路轨道不平顺预测模型研究与应用[J].铁道标准设计,2015,59(9):56-63.
[2]曲建军,高亮,田新宇,等.基于灰色理论的轨道几何状态中长期时变参数预测模型的研究[J].铁道学报,2010,32(2):55-59.
[3]许玉德,吴纪才.利用线性预测模型分析轨道不平顺发展[J].石家庄铁道学院学报,2005,18(1):6-9.
[4]陈宪麦,王澜,杨凤春,等.用于铁路轨道不平顺预测的综合因子法[J].中国铁道科学,2006,27(6):27-31.
[5]曲建军.基于提速线路TQI的轨道不平顺预测与辅助决策技术的研究[D].北京:北京交通大学,2011.
[6]韩晋,杨岳,陈峰,等.基于非等时距加权灰色模型与神经网络的轨道不平顺预测[J].铁道学报,2014,36(1):81-87.
[7] Akihito Kawaguchi,Miwa Masashi,Koichiro Terada. Actual data analysis of alignment irregularity growth and its prediction model[J].Quarterly Report of RTRI,2005,46(4):262-268.
[8]温丽华.灰色系统理论及其应用[D].哈尔滨:哈尔滨工程大学,2003.
[9]姜爱平,张启敏.非等间距近似非齐次指数序列的灰色建模方法及其优化[J].系统工程理论与实践,2014,34(12):3199-3203.
[10]党耀国,刘震,叶瞡.无偏非齐次灰色预测模型的直接建模法[J].控制与决策,2017,32(5):823-828.
[11] Suykens Jak,Vandewalle J. Least squares support vector machine classifiers[J]. Neural Processing Letters,1999,9(3):293-300.
[12] Suykens Jak,Vandewalle J. Recurrent least squares support vector machines[J]. IEEE Transactions on Circuits and Systems I:Fundamental Theory and Applications,2000,47(7):1109-1114.
[13]乐天晗,吴永军,吴湘华,等.优化GM(1,1)与SVM组合模型的路基冻胀预测应用[J].铁道科学与工程学报,2017,14(11):2345-2351.
[14] Seyedali Mirjalili,Andrew Lewis. The whale optimization algorithm[J]. Advances in Engineering Software,2016,95:51-67.
[15]牛培峰,吴志良,马云鹏,等.基于鲸鱼优化算法的汽轮机热耗率模型预测[J].化工学报,2017,68(3):1049-1057.
[16]曹净,丁文云,赵党书,等.基于PSO-LSSVM模型的基坑变形时间序列预测[J].控制工程,2015,22(3):475-480.
[17]冯胜洋,魏丽敏,郭志广.基于最小二乘支持向量机的高速铁路路基沉降预测[J].中国铁道科学,2012,33(6):6-10.
[18]曹净,丁文云,赵党书,等.基于LSSVM-ARMA模型的基坑变形时间序列预测[J].岩土力学,2014,35(S2):579-586.
[19]耿立艳,张天伟,赵鹏.基于灰色关联分析的LS-SVM铁路货运量预测[J].铁道学报,2012,34(3):1-6.
[20]张弦,王宏力.基于粒子群优化的最小二乘支持向量机在时间序列预测中的应用[J].中国机械工程,2011,22(21):2572-2576.
[21]宋晓华,祖丕娥,伊静,等.基于改进GM(1,1)和SVM的长期电量优化组合预测模型[J].中南大学学报(自然科学版),2012,43(5):1803-1807.
[22]唐杰明,刘俊勇,杨可,等.基于灰色模型和最小二乘支持向量机的电力短期负荷组合预测[J].电网技术,2009,33(3):63-68.
[2]曲建军,高亮,田新宇,等.基于灰色理论的轨道几何状态中长期时变参数预测模型的研究[J].铁道学报,2010,32(2):55-59.
[3]许玉德,吴纪才.利用线性预测模型分析轨道不平顺发展[J].石家庄铁道学院学报,2005,18(1):6-9.
[4]陈宪麦,王澜,杨凤春,等.用于铁路轨道不平顺预测的综合因子法[J].中国铁道科学,2006,27(6):27-31.
[5]曲建军.基于提速线路TQI的轨道不平顺预测与辅助决策技术的研究[D].北京:北京交通大学,2011.
[6]韩晋,杨岳,陈峰,等.基于非等时距加权灰色模型与神经网络的轨道不平顺预测[J].铁道学报,2014,36(1):81-87.
[7] Akihito Kawaguchi,Miwa Masashi,Koichiro Terada. Actual data analysis of alignment irregularity growth and its prediction model[J].Quarterly Report of RTRI,2005,46(4):262-268.
[8]温丽华.灰色系统理论及其应用[D].哈尔滨:哈尔滨工程大学,2003.
[9]姜爱平,张启敏.非等间距近似非齐次指数序列的灰色建模方法及其优化[J].系统工程理论与实践,2014,34(12):3199-3203.
[10]党耀国,刘震,叶瞡.无偏非齐次灰色预测模型的直接建模法[J].控制与决策,2017,32(5):823-828.
[11] Suykens Jak,Vandewalle J. Least squares support vector machine classifiers[J]. Neural Processing Letters,1999,9(3):293-300.
[12] Suykens Jak,Vandewalle J. Recurrent least squares support vector machines[J]. IEEE Transactions on Circuits and Systems I:Fundamental Theory and Applications,2000,47(7):1109-1114.
[13]乐天晗,吴永军,吴湘华,等.优化GM(1,1)与SVM组合模型的路基冻胀预测应用[J].铁道科学与工程学报,2017,14(11):2345-2351.
[14] Seyedali Mirjalili,Andrew Lewis. The whale optimization algorithm[J]. Advances in Engineering Software,2016,95:51-67.
[15]牛培峰,吴志良,马云鹏,等.基于鲸鱼优化算法的汽轮机热耗率模型预测[J].化工学报,2017,68(3):1049-1057.
[16]曹净,丁文云,赵党书,等.基于PSO-LSSVM模型的基坑变形时间序列预测[J].控制工程,2015,22(3):475-480.
[17]冯胜洋,魏丽敏,郭志广.基于最小二乘支持向量机的高速铁路路基沉降预测[J].中国铁道科学,2012,33(6):6-10.
[18]曹净,丁文云,赵党书,等.基于LSSVM-ARMA模型的基坑变形时间序列预测[J].岩土力学,2014,35(S2):579-586.
[19]耿立艳,张天伟,赵鹏.基于灰色关联分析的LS-SVM铁路货运量预测[J].铁道学报,2012,34(3):1-6.
[20]张弦,王宏力.基于粒子群优化的最小二乘支持向量机在时间序列预测中的应用[J].中国机械工程,2011,22(21):2572-2576.
[21]宋晓华,祖丕娥,伊静,等.基于改进GM(1,1)和SVM的长期电量优化组合预测模型[J].中南大学学报(自然科学版),2012,43(5):1803-1807.
[22]唐杰明,刘俊勇,杨可,等.基于灰色模型和最小二乘支持向量机的电力短期负荷组合预测[J].电网技术,2009,33(3):63-68.